The recommended treatment for HF in DM (symptomatic or to prevent HF hospitalization and/or death) is similar to treatment of HF in general and includes ACEis, β-blockers, MRAs, ARBs, and diuretics. Ivabradine or ARNI should be considered in the case of persistent symptoms and EF < 35%, and digoxin may be considered in patients with sinus rhythm and persistent symptoms. The mechanisms for clinical effects of these interventions are shown in Fig. 2. There is so far no evidence for a different treatment response in patients with or without DM in the large HF trials (n = 38,600 patients, Tables 1 and 2 and Fig. 3a, b), which means that the absolute benefit in patients with DM is greater due to increased risk of HF events. Side effects are also similar, except for increased risk of hyperkalemia with some agents blocking the RAAS system [3, 25, 76]. However, these data are, in lack of dedicated HF trials in T2DM patients, derived from subgroup analyses which have intrinsic limitations in terms of generalizability. Interpretation may therefore be challenging, in particular since the T2DM population often differs compared to the non-T2DM population in disease duration, metabolic control and vascular disease burden. Some HF studies furthermore report heterogeneity for the effect on HF hospitalizations and mortality with respect to geographical region [8, 77], likely explained by a multitude of factors including differences in the approach to diagnosis and etiology, availability of resources, and social and cultural circumstances [78].

Of note, to date, no treatments have proven to be effective in reducing morbidity and mortality in HFpEF, and guidelines recommend symptomatic treatment of fluid overload with diuretics in addition to controlling risk factors such as blood pressure and atrial fibrillation [3]. Since there also are fewer studies conducted in HFpEF as compared to HFrEF, recommendations of HF treatments are mainly based on evidence from studies in patients with HFrEF (Tables 1 and 2). The “CHARM-preserved” and “I-preserved” studies found no benefit on mortality from the ARBs candesartan and irbesartan in HFpEF and only a moderate benefit on hospitalization for HF [79, 80]. Approximately 27% of the study population in both trials had T2DM with no heterogeneity in the results [22]. The Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist (TOPCAT) trial explored the effect of spironolactone on a primary outcome of CV death, aborted cardiac arrest and hospitalization for HF in a population with EF ≥ 45% of which approximately 30% had DM, with no significant effects on the primary outcome [81]. One discussed explanation for the neutral effect was the regional differences observed with higher event rates and significant effect in patients included in the Americas and lower event rates and no treatment effect in those included in Russia/Georgia, potentially caused by different practice patterns and use of hospitalization [8]. Three other smaller pilot studies with digoxin, perindopril, and carvedilol in HFpEF also failed to show any beneficial impact on survival or HF hospitalizations [82‐84]. Sacubitril/valsartan (ARNI) is a new treatment option which combines an ARB with a neprilysin inhibitor and now is being studied in patients with symptomatic HFpEF (EF ≥ 45%) in the “Prospective comparison of ARNI with ARB Global Outcomes in heart failure with preserved ejection fraction” (PARAGON-HF) [https://​clinicaltrials.​gov/​ct2/​show/​NCT01920711?​term=​PARAGON&​rank=​4.] which will be informative for its use also in this patient group.

A dysregulated RAAS is a characteristic of both HF and T2DM, and RAAS inhibition is a recommended therapy in both conditions. Studies indicate that ACEis have a similar magnitude of effect on reducing mortality and HF hospitalization in populations with HF and prevalent T2DM as in those without T2DM [17, 85]. A post hoc analysis of the Studies of Left Ventricular Dysfunction (SOLVD) treatment and prevention studies illustrated this nicely showing that there were no interactions by the number and type of non-cardiac comorbidities (including T2DM) regarding treatment effects of enalapril on outcomes [86] (Table 1 and Fig. 3a, b). Enalapril furthermore slowed LV remodeling and the development of, and hospitalization for, HF when given to asymptomatic patients with LV dysfunction (EF ≤ 35%) [87].

MRAs reduce symptoms and mortality in both mild and severe symptomatic HF, with consistent effects also in the DM subgroups [26, 88] (Table 1, Fig. 3a, b), and are recommended as add-on therapy to β-blocker and an ACEi/ARB if persisting HF symptoms.

ARBs are less investigated, but have proven similar effects on mortality and HF hospitalization as ACEis when given as an alternative in ACEi-intolerant patients [89] (Table 2, Fig. 3a, b). They are therefore recommended as an alternative to, but not as an add-on to, ACEis, since they have not shown consistently to provide synergistic effects in reducing mortality [22, 75, 90].

Due to compensatory renin activation under treatment with ACEis or ARBs, it was postulated that a dual RAAS blockade with a renin inhibitor in combination, would improve cardiorenal outcomes in a T2DM population. In the Aliskiren Trial in Type 2 Diabetes Using Cardiorenal Endpoints (ALTITUDE), a study of patients with T2DM and kidney disease of whom 11% had a history of chronic HF, aliskiren increased the risk of side effects (in particular hyperkalemia) without benefit on outcomes, and hence renin inhibition is not recommended for patients with T2DM on ACEi or ARBs [76].

β-Blockers are, together with ACEis and MRAs, the cornerstones in the treatment of HF. Several large randomized clinical studies have reported reduced rates of mortality and hospitalization for HF with β-blockers in patients with HFrEF [20, 72, 73, 91], with no significant heterogeneity according to diabetes status (Table 1, Fig. 3a, b). A meta-analysis confirmed these findings in T2DM subjects [92]. Although carvedilol in one study was suggested to improve survival as compared to metoprolol, both in the overall, as well as in the patients with prevalent T2DM [93, 94], guidelines do not suggest a preferred β-blocker [1, 3]. β-Blockers are however still, potentially, underutilized in T2DM, perhaps due to fear of side effects, in particular blunting of symptoms of hypoglycemia. Post hoc analyses of the large β-blocker trials have furthermore reported some regional differences with smaller survival benefit in patients included in North America than in the rest of the world [77, 95]; however, these findings remain to be confirmed in prospective studies powered to explore geographical variations in treatment response.

Diuretics relieve symptoms of fluid retention in HF [96], but the effect on mortality is not established. Diuretics, loop or thiazide, are indicated when HF is decompensated with symptoms of fluid retention.

Digoxin/cardiac glycosides have been widely used in HF patients, both in the setting of sinus rhythm and atrial fibrillation, with or without prevalent DM. The Digitalis Investigation Group (DIG) trial included patients with HF with EF ≤ 45% and sinus rhythm, of which 28% had prevalent DM (Table 2) [97]. The overall study results showed no impact on mortality, but a reduction in the hospitalization for HF [18], which was confirmed in a later meta-analysis [98]. Subgroup analyses from the DIG trial showed that the effect tended to be more pronounced in the subgroups with the most severe HF (EF < 25%, NYHA class III or IV) [18, 99]. The subgroup by DM analyses were recently published and showed no significant interactions for any of the outcomes [19]. Of note is that very few patients used β-blockers and no patients used MRAs in the DIG trial; thus, the effect on outcomes from digoxin used on top of current standard of care is not known.

Sacubitril inhibits neprilysin endopeptidase, blocking the catabolism of natriuretic peptides and other vasoactive peptides and thereby increasing their bioavailability (Fig. 2). The combination of valsartan and sacubitril was studied in the Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure (PARADIGM-HF) Trial involving HFrEF patients only [27]. The trial reported a reduced risk for CV death and hospitalization for HF for ARNI when compared to treatment with enalapril in symptomatic patients with HFrEF (Table 2) [27]. The results were consistent across subgroups, including the subgroup with T2DM comprising approximately 35% of the study population (Table 2 and Fig. 3a, b). Sacubitril/valsartan is now recommended in the HF guidelines of the ESC as a replacement for an ACEi to further reduce the risk of HF hospitalization and death in ambulatory patients with HFrEF who remain symptomatic despite optimal treatment with an ACEi, a β-blocker, and an MRA [3].

Ivabradine, an inhibitor of the cardiac pacemaker current I_f, reduces heart rate thereby reducing the cardiac work burden (Fig. 2). The Systolic Heart Failure Treatment With the I_f Inhibitor Ivabradine Trial (SHIFT) included HFrEF patients with sinus rhythm and a heart rate of at least 70 beats per minute (bpm) and demonstrated significant improvements in both the composite endpoint of HF hospitalization and CV death, and for HF hospitalization alone, with a similar magnitude of effect in those with prevalent DM [23] (Table 2 and Fig. 3a, b). Ivabradine is recommended if HF symptoms persist despite treatment with a β-blocker, ACEi, and MRA (in patients with sinus rhythm > 70 bpm).

Metformin is the recommended first-line blood glucose-lowering treatment in T2DM patients with HF [3]. A previous contraindication for its use in HF, due to concerns for lactic acidosis, was removed by the FDA in 2007 [106] following retrospective studies reporting improved outcomes with lower re-admission and mortality rates as compared to other glucose-lowering treatments in patients with T2DM and HF [107‐109]. A later, larger systematic review supported this conclusion [110]. No RCTs of metformin indicate, however, a role in the prevention of HF or HF outcomes.

In patients with newly diagnosed T2DM enrolled in the UKPDS there was no increased risk of HF associated with the use of sulphonylurea [111], a finding also seen in the ADVANCE trial [112]. However, some retrospective studies have indicated that second generation sulphonylureas might be associated with 18–30% increased risk of HF as compared to metformin [113, 114]. One Canadian retrospective study, based on the Saskatchewan Health records, reported similar results (i.e., increased HF admission rates associated with sulphonylurea as compared to metformin), but when background characteristics (e.g., history of coronary heart disease, use of CV medication) were adjusted for, the difference was no longer significant [115]. The safety of sulphonylurea in HF populations with DM is thus not fully established and ongoing trials like the Italian Thiazolidinediones or Sulfonylureas and Cardiovascular Accidents Intervention Trial (TOSCA-IT) [116]), CARdiovascular Outcome Trial of LINAgliptin Versus Glimepiride in Type 2 Diabetes (CAROLINA®) trial [117, 118] and The Glycemia Reduction Approaches in Diabetes: A Comparative Effectiveness Study (GRADE) [119] might provide further insights in these matters.

With the expanding armamentarium of non-insulin therapies for T2DM, insulin initiation typically occurs late in the T2DM disease trajectory [120]. Consequently, we observe a general pattern of a more deleterious cardiometabolic risk profile among patients initiating insulin [121] with a corresponding relative higher occurrence of CV events as compared to non-insulin users [122]. Mechanistically, since insulin therapy can induce weight gain [123], sodium retention, and fluid retention [124] and is being discussed to have some other vascular detrimental effects [125], it has been postulated that such therapy in T2DM potentially could worsen outcomes, particularly in a HF setting. However, in well powered RCTs assessing the effect on CV outcomes, including HF outcomes, from insulin-based intensive versus conventional glucose lowering, neither beneficial nor harmful effects on HF were observed, i.e., in the Outcome Reduction with an Initial Glargine Intervention (ORIGIN) trial where HF hospitalization HR was 0.91 (95% CI 0.87–1.05]) [126] and UKPDS where HF RR was 0.78 (95% CI 0.39, 1.55) [111]). Recently, a consistent result in the long-term follow-up of ORIGIN was also reported (HR 1.03 (95% CI 0.97, 1.10]) [127]. There are however no trials designed to test effects on CV outcomes, including HF, of insulin-treatment with the intent of achieving glycemic equipoise between treatment arms.

Thiazolidinediones (TZDs) are insulin sensitizing drugs known to cause fluid retention. The class includes rosiglitazone and pioglitazone, which both in dedicated outcome trials were associated with increased risk of HF hospitalizations, even though patients with a previous history of HF (NYHA class II-IV) were excluded; the Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD) study reported a HR of 2.10 (95% CI 1.35, 3.27) [128], and the PROspective pioglitAzone Clinical Trial In macroVascular Events (PROactive) a HR of 1.41 (95% CI 1.10, 1.80) [29] (Table 3, fig. 4a and b). Although the drugs might modulate the risk of MI differently [129], a meta-analysis indicated that they increase the risk of HF to a similar extent [130]. One study aiming to understand the cardiac dynamics with TZD therapy found that rosiglitazone significantly increased left ventricular end-diastolic volume [131]. The use of TZDs is contraindicated in patients with known or prior HF with functional class NYHA I-IV and if used, patients with known HF risk factors should be monitored for HF symptoms such as oedema and weight gain.

The drug class of DPP4 inhibitors has accumulated solid evidence from RCTs on risk of HF, with three large CV outcome trials (Table 3) recently completed and reported, all with hospitalization for HF as a pre-specified secondary or exploratory outcome [36, 37, 40] (Table 3). The Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR-TIMI) 53 trial compared saxagliptin and placebo on top of standard care in T2DM patients either with established CVD or at risk for CVD and found no difference in the primary outcome (a composite of CV death, MI, or ischemic stroke). However, patients in the saxagliptin group were at higher risk of being admitted to the hospital for HF (HR 1.27, 95% CI 1.07–1.51, p = 0.007) than were those in the placebo group (Fig. 4a). Interestingly, but not surprisingly, the increased risk was associated with increased levels of NT-proBNP and an estimated GFR ≤ 60 ml/min/1.73 m² at baseline, regardless of treatment allocation [37], and in absolute term, patients with prevalent HF at baseline had a higher risk (Fig. 4b). However, the incremental HF signal with saxagliptin was most prominent in those without prevalent HF (HR 1.32 (95% CI 1.04, 1.66) vs HR 1.21 (95% CI 0.99, 1.58)) (Table 3 and Fig. 4b). The Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care (EXAMINE) study compared alogliptin to placebo in T2DM patients with a recent acute coronary syndrome [36], of which 28% had reported HF at inclusion. There was no difference between the groups in the risk of the primary outcome (composite of CV death, non-fatal MI, or non-fatal stroke), but there was a numerically increased risk of hospitalization for HF associated with alogliptin (HR 1.19 (95% CI 0.89, 1.58)) (Table 3). The risk of the primary outcome was consistent in the subpopulation with established HF at inclusion; however, as observed in SAVOR-TIMI53, a relative higher risk for HF hospitalization with alogliptin therapy was seen in patients without prevalent HF (HR 1.76 (95% CI 1.07, 2.90)) as compared to those with HF (HR 1.00 (95% CI 0.71, 1.42)) (Fig. 4b). The Trial Evaluating Cardiovascular Outcomes With Sitagliptin (TECOS) investigated the effect of sitagliptin vs placebo in T2DM patients with established CVD, with neutral effect on the primary outcome (CV death, non-fatal MI, non-fatal stroke, or hospitalization for unstable angina) [40]. There was no difference in the risk for hospitalization for HF between the treatment groups (HR 1.00 (95% CI, 0.83 to 1.20)), regardless of HF status at baseline. A mechanistic study exploring the impact of vildagliptin versus placebo on systolic function as measured by EF with echocardiography in 254 T2DM patients with HF in NYHA class I-III found no differences in change in EF after 1 year (primary outcome), but a larger increase in both LV end-diastolic and end-systolic volumes was seen with vildagliptin [132], an effect of similar magnitude as observed with rosiglitazone therapy [131]. Finally, although no dedicated outcome study has been reported yet for linagliptin, a meta-analysis of pooled registration studies indicated no increased risk of hospitalization for HF [133].

Thus, the published trials so far with DPP4 inhibitors suggest that the increased risk of HF seen with certain class members does not represent a class effect. In the coming years two CV outcome trials with linagliptin will be reported, i.e., the CAROLINA® trial [117, 118] (linagliptin vs sulphonylurea) in 2019 and the CArdiovascular Safety & Renal Microvascular outcomE study with LINAgliptin (CARMELINA®) (linagliptin vs placebo) in 2018 [https://​clinicaltrials.​gov/​ct2/​show/​NCT01897532?​term=​CARMELINA&​rank=​1.].

GLP-1 receptor agonists are indicated to reduce glucose in T2DM and belong to the incretin class of drugs. Apart from glucose-lowering effects, they also have a number of non-glycemic effects including reducing appetite (and inducing weight loss), modestly reducing BP and increasing pulse rate. Two sufficiently powered CV outcomes trials within this drug class have thus far reported; the Evaluation of LIXisenatide in Acute Coronary Syndrome (ELIXA) trial and the Liraglutide Effect and Action in Diabetes: Evaluation of cardiovascular outcome Results—a long term evaluation (LEADER) trial [31, 33]. The ELIXA trial (Table 3) included 6068 T2DM patients with recent acute coronary syndrome (< 180 days) and reported a neutral effect on both the primary outcome (composite of CV death, non-fatal stroke, non-fatal MI and hospitalization for unstable angina) and on hospitalization for HF (HR 0.96 (95% CI, 0.75 to 1.23)) (Fig. 4a, b) [31]. In this large trial, only a slightly increased heart rate was observed with lixisenatide, with mean 0.4 bpm (95% CI 0.1, 0.6), a result potentially influenced by the high use of β-blockers in the study population (85 and 84% at baseline in the placebo and lixisenatide groups, respectively). The LEADER trial (Table 3) followed 9340 T2DM patients with established CVD (approximately 80%) or more than one CV risk factor (approximately 20%) for a median time of 3.8 years, and reported significantly decreased risk for the primary outcome (composite of CV death, non-fatal stroke and non-fatal MI) with liraglutide as compared to placebo (HR 0.87 (95% CI 0.78, 0.97)). This result was driven by a 22% relative reduction in CV death (HR 0.78 (95% CI 0.66, 0.93)), whereas non-fatal MI and non-fatal stroke were not significantly affected. Fourteen % of the study population had prevalent HF, and the drug had no impact on hospitalization for HF (HR 0.87 (95% CI 0.73, 1.05) Fig. 4a, b)) despite the slightly increased heart rate seen with liraglutide of 3 bpm relative to placebo. Lately, smaller studies have suggested potentially adverse effects on cardiac function of liraglutide: the Effects of Liraglutide on Clinical Stability Among Patients With Advanced Heart Failure and Reduced Ejection Fraction: A Randomized Clinical Trial (FIGHT) trial [134] randomized 300 adults (60% with T2DM) with acute decompensated HFrEF to 1.8 μg liraglutide or placebo. After 6 months, there was a numerically increased risk for death and hospitalization for HF (which were parts of a hierarchical primary endpoint together with NT-proBNP levels) with liraglutide, and this finding was accompanied by increased LV diastolic and systolic volumes. In this study, there was no difference in the change of heart rate, but there were more cases of arrhythmia with liraglutide reported as safety events (17 vs 11% in liraglutide and placebo). These findings were in line with the Effect of Liraglutide on Left Ventricular Function in Chronic Heart Failure Patients With and Without Type 2 Diabetes Mellitus (LIVE) study [135] including 241 patients with chronic HFrEF (approximately 30% had T2DM) where no impact on systolic function by echocardiography was seen, but a significantly increased heart rate with 6 bpm with liraglutide vs 1 bpm with placebo, p < 0.001. The mechanisms behind the increased heart rate and further effects on myocardial function with GLP-1 receptor analogues remain to be elucidated [136‐138].

SGLT-2 inhibitors reduce glucose reuptake in the kidneys by inhibiting the SGLT-2 transport protein thereby causing glucosuria, urinary caloric loss, and volume loss (osmotic diuresis, transient natriuresis). Within this class of drugs, results from two sufficiently powered CV outcome trials have been reported, i.e., EMPA-REG OUTCOME® testing empagliflozin versus placebo and the CANVAS Program (CANagliflozin cardioVascular Assessment Study), testing canagliflozin versus placebo (Table 3). The EMPA-REG OUTCOME trial randomized and treated 7020 T2DM patients with established CV disease, of which 10% had prevalent HF, to assess CV safety of empagliflozin given on top of standard of care [41]. The primary outcome (CV death, non-fatal MI, and non-fatal stroke) was significantly reduced by 14%, driven by a reduction in CV death by 38%. In both the placebo and empagliflozin groups, the most frequent modes of CV death were sudden death, death from HF, and presumed CV death (death of unknown cause), and all categories of CV death contributed to the risk reduction with empagliflozin. Notably, hospitalization for HF was reduced by 35% (HR 0.65 (95% CI 0.50, 0.85)) with 4.1 and 2.7% of patients in placebo and empagliflozin groups being hospitalized with HF (Table 3, Fig. 4a, b), as was the composite of hospitalization for HF and CV death (HR 0.66 (0.55, 0.79)) and time to introduction of loop diuretics (0.62 (95% CI 0.53–0.73). Subgroup analyses revealed no significant heterogeneity with regards to baseline kidney function, CV medication used or prevalent HF [42]. Recent guidelines on the diagnosis and treatment of HF issued by the ESC recommend that the use of empagliflozin be considered (class IIa recommendation) in patients with T2DM to prevent or delay the onset of HF and prolong life [3], and FDA also recently approved empagliflozin to reduce the risk of CV death in adults with T2DM and established CV disease.

The CANVAS Program, combining data from two independent trials (the CANVAS trial and the CANVAS-R trial) [139], included patients with established CVD or being at high CV risk, of which approximately 14% had HF at baseline [105]. In this program, when compared to placebo, canagliflozin reduced the primary outcome (CV death, non-fatal MI, non-fatal stroke) by 14% (HR 0.86 (0.75–0.97) [105], but without any significant effect on CV or all-cause death. Hospitalization for HF was however reduced (HR 0.67 95% CI 0.52–0.87), with similar magnitude as with empagliflozin with 2.1 and 2.8% of the patients on canagliflozin and placebo, respectively, being hospitalized for HF during the program. At variance with the safety findings of empagliflozin, canagliflozin was associated with a significant increased risk for lower leg amputation and bone fractures [105].

There are several mechanisms potentially explaining the benefits of empagliflozin and canagliflozin on HF hospitalizations, one being the reduction in blood pressure, arterial stiffness, double product (also known as rate pressure product) and pre-load without any compensatory increase in heart rate [140, 141]. A reduction in weight and visceral fat may also play a role [142, 143], as may the reduction in uric acid and improved energy utilization, and for empagliflozin, an increase in hematocrit [41, 144]. Data from the EMPA-REG OUTCOME® trial and the CANVAS Program also revealed significantly decreased progression in nephropathy and risk of adverse renal outcomes (dialysis, doubling of s-creatinine, renal transplant) [145] which is likely due to the reduction in glomerular hypertension and restored tubule-glomerular feedback [146]. Thus, several mechanisms may contribute to the effects of empagliflozin and canagliflozin on HF outcomes [147‐149]. CV outcomes trials with the other SGLT-2 inhibitors are due to be published within the next few years, all with hospitalization for HF as a pre-specified secondary outcome [https://​clinicaltrials.​gov/​ct2/​show/​NCT01730534?​term=​declare+timi&​rank=​1., https://​clinicaltrials.​gov/​ct2/​show/​NCT01986881?​term=​vertis&​rank=​1.].

In a meta-analysis of phase 2 and 3 trials with dapagliflozin there was no sign of increased risk of HF hospitalization as compared to placebo or comparator (HR 0.36 (95% CI 0.16, 0.84), but the analysis was based only on 26 events [150].